Algorithm 1: K-medoids for uncertain data using a probabilistic distance measure in feature space.

1.  Input: $\mathit{n}$: The number of objects in cluster k, $\mathit{K}$: The number of clusters, iter = 0; 2.  Randomly select the cluster medoids ${\mathit{C}}^{\left(\mathbf{0}\right)}=\{{\mathit{c}}_{\mathbf{1}}^{\left(\mathbf{0}\right) },\dots ,{\mathit{c}}_{\mathit{K}}^{\left(\mathbf{0}\right)}\}$ obtained from the initial clusters 3.  Initialize 4.  $\mathit{C}\mathit{V}\mathit{I}\mathit{s}=\left\{\mathit{c}\mathit{v}{\mathit{i}}^{\left(\mathbf{1}\right)}, \dots , \mathit{c}\mathit{v}{\mathit{i}}^{\left(\mathit{K}\right)}\right\}$ obtained UOSDU, UOSCH, UFSDU, and UFSCH 5.  Repeat 6.  for $k=2$ to $K$ 7.    ${\mathit{c}}_k^{\left(old\right)}={\mathit{c}}_k^{\left(0\right)}$; ${\mathit{c}}_k^{\left(new\right)}=0$ 8.    Compute the new medoids: 9.    while $c_k^{\left(old\right)} \ne c_k^{\left(new\right)}$  10.      $p=\underset{1\le i\le n}{\underbrace{argmin}}{\displaystyle \sum }_{j=1}^{k}KP{D}_{Bhatt}({\mathit{x}}_i, {\mathit{c}}_{jk}^{ })$, where $j$ is an index of cluster medoid in ${\mathit{c}}_k^{ }$  11.      ${\mathit{c}}_k^{\left(new\right)}={\mathit{x}}_p$  12.    end 13.    Calculate the $\mathit{c}\mathit{v}{\mathit{i}}^{\left(k\right)}$ using Equations (1), (2), (17), and (18). 14.  end 15.  iter = iter + 1 16. Until ($\mathrm{iter}$ = Maxiter)